Application of Geographically Weighted Regression to Improve Grain Yield Prediction from Unmanned Aerial System Imagery